Canonical correlation forests are an extension of decision trees with improved performance, particularly on datasets with correlated features. Paper is available here:
http://arxiv.org/abs/1507.05444 and reference implementation (in Matlab) is here: https://bitbucket.org/twgr/ccf Is there any interest in adding this? Since there's already an implementation of CCA, it seems to me (perhaps naively) straightforward. I'm not sure I know enough about the sklearn implementation of decision trees or CCA to add this, but I'm hoping there's someone on this list with the right skills & interest. -- Scott Turner
------------------------------------------------------------------------------
_______________________________________________ Scikit-learn-general mailing list Scikit-learn-general@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/scikit-learn-general
